On Saturday 14 November 2015, Nick Koretsky wrote:
> On Sat, 14 Nov 2015 10:14:10 -0500
>
> Gene Heskett <gheskett@...> wrote:
> > Greetings;
> >
> > Twice in the last 2 weeks I have been forced to reboot the machine, as it
> > had a locked up, unkillable session of kmail showing in htop.  I had hit
> > the list reply button, only to be greeted by the spinning dumbell mouse
> > cursor, which would spin forever.  The machine was NOT stuck in a loop
> > someplace burning up cpu's, nothing, not even kmail was showing any
> > great cpu activity, but the process itself was unkillable.
> >
> > This time I was attempting to teach Bret Busby how to construct a sig
> > separator, the \r-- \r sequence and it was one of his reply messages I
> > was reading & tried to reply to when it locked up.
> >
> > This reboot took around 30 minutes because /amandatapes-1T, a terabyte
> > drive, had not been fsck'd in 181 days, and fscking a terabyte drive,
> > even if no errors other than severe fragmentation (62%) exist is a 30 to
> > 45 minute process.  Amanda, the backup program, /thinks/ the drive is a
> > 30 tape, changer equiped library. (except its 10,0000% more dependable
> > that such a mechanical monster would be, and 2% of the cost) I'd hate to
> > think how long it will take to do a 2 terabyte drive as there is now 4
> > of them here at the Heskett mini-ranchette. Its difficult to buy
> > anything smaller.
> >
> > I deleted the last 4 messages from him from ~/Mail/kubuntu-users/cur
> > before I restarted kmail, knowing I'd be going thru the same hung
> > machine procedure if I didn't.  Last time I made 3 reboots trying, and
> > finally just cleaned out the folder (a  different one) before trying a
> > 4th reboot & kmail restart.  That, like this, complained about the index
> > file being corrupted on the launch of kmail but normal operation was
> > able to be resumed.
> >
> > Has anyone a clue just what gmail is doing now?  Or how to fix kmail so
> > it tolerates a message posted from their (IMNSHO) busted webmail app?
> >
> > This is R14.0.0.2 I believe.  Last updated from Slavics? server a day
> > back.
> > Looks like I'll need to reboot again, I have a net caused ooops in the
> > log, and my local network is dead although I can ping yahell.
>
> Thats weird. Generally speaking the process cannot become unkillable due
> to error in a program itself, thats only happens if it stuck in the system
> call. So i suspect problems with disk/fs or maybe a faulty network
> card driver.

That makes sense. I'd do an HD check with the maker's tool CD. 
Make sure to run a SMART check as, silent HD hang ups are often electronic 
rather than mechanical.

If it's a desktop, swap out the nic.

Good luck

Kate